Objective prior distributions for Jolly-Seber models of zero-augmented data.

Area of Science:

• Ecology
• Statistics
• Population Dynamics

Background:

• Jolly-Seber (JS) models are statistical tools for estimating population dynamics from capture-recapture data.
• State-space versions of JS models handle zero-augmented data, including many unobserved individuals.
• Standard uniform priors in JS models can bias population size (N) estimates when N is large relative to observed data.

Purpose of the Study:

• To derive a novel class of prior distributions for JS model recruitment parameters.
• To enable objective prior specifications for population size (N).
• To allow incorporation of prior knowledge while maintaining objective priors on N.

Main Methods:

• Derivation of a new class of prior distributions for JS model recruitment parameters.
• Development of methods to specify objective priors for population size (N).
• Analysis of simulated and real capture-recapture data.

Main Results:

• The derived priors include discrete-uniform and improper scale priors as special cases.
• These priors mitigate the influence of uniform priors on N and other parameter posteriors.
• Demonstrated inferential benefits using simulated and real datasets.

Conclusions:

• The new class of priors provides a more robust framework for estimating population dynamics using JS models.
• These priors enhance the objectivity and flexibility of capture-recapture data analysis.
• Identified conditions under which these priors yield the greatest inferential advantages.